Dynamic barriers show distinct impacts at fixed occupancy. (A) Biophysical measurements of residence times for CTCF (red) and cohesin (orange) in mouse cells. Measurements for individual cell lines or acquisition rates shown as separate points. Shapes indicate SPT (triangle) or FRAP (circle). See Supplemental Table S1 for details. (B) Top: Illustration of CTCF as a dynamic barrier to loop extrusion as a function of time, where CTCF sites are bound with timescale τb and extruders have lifetime τE. Bottom: Lattice implementation. Extruder position at previous time step shown as a light arch. If a barrier becomes unbound (middle), an extruder blocked at this site can continue extruding. Note: CTCF can rebind when the barrier is inside of an extruded loop. (C) Representative traces for the state of a single barrier (bound or unbound) versus time, for three different regimes of barrier bound time versus extruder lifetime at fixed barrier occupancy. (D) Snapshots of extruder positions in different cells generated by the dynamic barriers model for a particular choice of barrier dynamics, where arcs indicate positions connected by the left and right extruder legs. To generate in silico ChIP-seq tracks, extruder positions are recorded for both legs from 500,000 timepoints, representing a population of cells, and averaged. (E) In silico ChIP-seq for extruders and barriers across a 1.25-Mb region for three different regimes of extruder versus barrier lifetimes, yet fixed barrier occupancy (0.9). Displayed barrier bound times τb (4 sec, 1350 sec, 4050 sec) range from much less to much greater than extruder lifetime (τe = 1350 sec). Barrier ChIP-seq displays equal height peaks, because occupancies are fixed. Extruder ChIP-seq displays strong peaks for long barrier bound times (τb > τe) and very weak peaks for transient barriers (τb ≪ τe). Note that variation in extruder peak heights comes from differential spacing between barriers, as all barriers here have identical parameters. (F) Extruder positions from 1D lattice simulations are used as input to 3D polymer simulations. Ensembles of 3D conformations are generated to produce contact maps. (G) In silico contact maps for the same simulated region and parameters as E, binned to 2.5-kb resolution. Barriers with similar (τb ∼ τE) or longer (τb > τE) bound times than the extruder lifetime display TADs and dots. In contrast, these patterns vanish with transient barriers (τb ≪ τE), even at the same occupancy.
